←back to thread

Google’s FLoC Is a Terrible Idea

(www.eff.org)
604 points wyldfire | 1 comments | 04 Mar 21 15:59 UTC | HN request time: 0.21s | source
Show context
brofallon ◴[04 Mar 21 19:23 UTC] No.26346918[source]▶
I'm wondering how hard it would be to reverse-engineer the FLoC algorithm that assigns ids based on browsing history.. could one just have a bunch of headless browsers randomly visit sites, and compute the FLoC ids periodically to see what types of sites end up producing which ids? This seems important since being assigned a group ID that includes a bunch of people might not be so bad, but (as the article suggests) if its well known which web sites are included in the group, that's a more disturbing story
replies(1): >>26349605 #
1. DreadY2K ◴[04 Mar 21 22:48 UTC] No.26349605[source]▶
If you run an advertising company, you could probably just pay people a small amount of money to get their browsing history and their FLoC id, and enough people would take you up on that offer that you'd get that data (and maybe also demographics), without having to do any work to reverse-engineer.